In evolved Universal terrestrial radio access network (EUTRAN) both real time IP multimedia Subsystem (IMS) and non-real time services such as Internet access may be provided using the Evolved Packet Core (EPC). EUTRAN coverage will initially be limited and thus existing voice data and data calls must be possible to hand over to e.g. UTRAN and GERAN. UTRAN/GERAN network deployments usually rely on legacy a circuit switched (CS) Core Network for handling voice calls while e.g. non-real time data services may be provided over a GPRS Core network. The deployment of parallel CS infrastructure in the market to support UTRAN and GERAN voice calls is expected to continue to exist for a long time.
To be able to handle this scenario an optional functionality called Single Radio Voice Call Continuity (SRVCC) has been defined in 3GPP release 8 in TS 23.2016 and TS 23.216.
TS 23.216 describes stage 2 functionality for SRVCC between EUTRAN access and 3GPP 1xCS and between EUTRAN access and 3GPP UTRAN/GERAN accesses and between UTRAN (HSPA) access and 3GPP UTRAN/GERAN accesses for circuit switched calls that are anchored in the IMS.
GPRS includes a preservation function, which allows an active PDP context associated with released Radio access bearers (RAB) to be preserved in the core network (CN) when a UE goes out of coverage. The RABs can then be re-established at a later stage when a UE comes back into coverage. PDP contexts associated with a guaranteed bit rate (GBR), Quality of Service (QoS) are preserved when the maximum bit rate (MBR) for uplink and downlink is set to 0 kbit/s.
When SRVCC is performed from EUTRAN to UTRAN/GERAN with dual transfer mode (DTM), packets switched (PS) handover support, as described in TS 23.216 section 6.2.2.2, there exists a potential race condition that may result in the voice call being dropped prematurely. A mobility management entity (MME) differentiates or “splits” the voice bearers from all other PS bearers and initiates their relocation towards a MSC server and an SGSN. After some time the MSC server, or MGW, initiates a session transfer towards IP multimedia Subsystem (IMS), the IMS, or a P-CSCF, will update service session information in a PCRF by removing affected media components using an Rx interface—step “A”.
In parallel to this procedure, the MSC server/MGW responds to the handover request to the source MME, which in turn initiates a handover command to the UE. The UE then tunes into UTRAN/GERAN and the handover is completed. Finally, an S4-SGSN updates bearers with a serving gateway (SGW). The SGW forwards the request to a Packet data network gateway (PGW). Since the voice bearers can not be established by the S4-SGSN, the S4-SGSN will release those bearers by triggering a bearer context de-activation towards the SGW and PGW. When the PGW receives an indication that the voice bearers have been de-activated it will report this to the PCRF over a Gx interface—step “B”.
The PCRF will now be able to match the information received over Rx with the information over Gx and take the necessary action, e.g. to remove the installed PCC rule in the PGW.
However, in these procedures step “B” may sometimes occur before step “A”. If this is the case the PCRF will have to assume that the voice bearer has been lost and report this to the P-CSCF. TS23.228 describes that the P-CSCF may actually initiate release of the whole IMS session in case of a bearer related to the IMS is lost.
The preservation function as described in TS 23.060 is not supported in EUTRAN EPS, but it could be added to EPS as well; however, then the whole QoS information element (IE) would have to be sent when indicating that the bearer has been preserved—HO to PS. Allowing the QoS IE to be sent from the MME/S4-SGSN will eventually lead to a risk of allowing QoS updates from MME/S4-SGSN which adds a considerable amount of complexity in the PGW.